_ Git - youtube-dl/blob - youtube_dl/extractor/aparat.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     ExtractorError,
   9     HEADRequest,
  10 )
  11
  12
  13 class AparatIE(InfoExtractor):
  14     _VALID_URL = r'^https?://(?:www\.)?aparat\.com/(?:v/|video/video/embed/videohash/)(?P<id>[a-zA-Z0-9]+)'
  15
  16     _TEST = {
  17         'url': 'http://www.aparat.com/v/wP8On',
  18         'md5': '6714e0af7e0d875c5a39c4dc4ab46ad1',
  19         'info_dict': {
  20             'id': 'wP8On',
  21             'ext': 'mp4',
  22             'title': 'تیم گلکسی 11 - زومیت',
  23         },
  24         # 'skip': 'Extremely unreliable',
  25     }
  26
  27     def _real_extract(self, url):
  28         video_id = self._match_id(url)
  29
  30         # Note: There is an easier-to-parse configuration at
  31         # http://www.aparat.com/video/video/config/videohash/%video_id
  32         # but the URL in there does not work
  33         embed_url = ('http://www.aparat.com/video/video/embed/videohash/' +
  34                      video_id + '/vt/frame')
  35         webpage = self._download_webpage(embed_url, video_id)
  36
  37         video_urls = [video_url.replace('\\/', '/') for video_url in re.findall(
  38             r'(?:fileList\[[0-9]+\]\s*=|"file"\s*:)\s*"([^"]+)"', webpage)]
  39         for i, video_url in enumerate(video_urls):
  40             req = HEADRequest(video_url)
  41             res = self._request_webpage(
  42                 req, video_id, note='Testing video URL %d' % i, errnote=False)
  43             if res:
  44                 break
  45         else:
  46             raise ExtractorError('No working video URLs found')
  47
  48         title = self._search_regex(r'\s+title:\s*"([^"]+)"', webpage, 'title')
  49         thumbnail = self._search_regex(
  50             r'image:\s*"([^"]+)"', webpage, 'thumbnail', fatal=False)
  51
  52         return {
  53             'id': video_id,
  54             'title': title,
  55             'url': video_url,
  56             'ext': 'mp4',
  57             'thumbnail': thumbnail,
  58         }